stop_all() {
	systemctl stop systemd-udevd-control.socket
	systemctl stop systemd-udevd-kernel.socket
	systemctl stop systemd-udevd.service
	systemctl stop systemd-udev-trigger.service
	systemctl stop systemd-udev-settle.service
}

stop_all

/usr/lib/systemd/systemd-udevd --daemon
udevadm trigger --action=add --type=devices
echo $?

#sleep 1
#udevadm settle

udevadm control --exit
echo $?

# If you stop the udevd sockets after the service, you might see:
# "Job for systemd-udevd.service canceled."

# If you do not stop the sockets, you might also see the message,
# and the job (applying a udev rule) will be finished after a long
# time by the udevd spawned

# If you do not add `sleep 1` after the trigger to give udevd
# extra time to finish its job, it'll take a long time to stop
# but still not finishing its job

# The settle service/command MIGHT work here, NOT because it
# has a mechanism to make sure everything is done (maybe it
# is expected to but it does not), BUT only because the extra
# "overhead" it brings, so it might not work if it is "faster"
# than `sleep 1` in certain case.

# The question is, can there be something more reliable than
# `sleep 1`?